BEIS 2016 - Workshop on Benchmark and Evaluation of Intelligent Surveillance (BEIS)

Smart/Intelligent video surveillance technology plays the central role in the emerging smart city systems. Recent advances in computer vision, such as deep learning, multi-modal analysis, large-scale spatio-temporal analysis, have shown great potential for some high level understanding tasks in smart video surveillance application. These include high performance human/object detection and tracking, cross camera human identification and re-identification, and action/activity/event detection. Based on these novel techniques, smart surveillance applications such as “person of interest tracing” and “fall/fighting detection” can be realized. Intelligent surveillance requires seamless integration of multiple technique building blocks including: human detection, tracking, re-identification, and action/event recognition, and these tasks have inter-connected nature. For instance, good human/object detection and tracking results are beneficial for event detection. This strongly requires joint modeling of various visual understanding tasks as well as joint evaluation of algorithmic performances. However, current benchmark datasets on video surveillance mainly contains separate tasks, and most of these existing datasets are constructed using self-collected video and images, not the real on-using surveillance data. These simulated datasets cannot evaluate the intelligent algorithms perfectly. Based on a newly collected multi-task multi-camera on-using surveillance dataset as well as evaluation metrics, this workshop aims at bringing together cutting-edge researches in the field of multi-task aware, multi-modal oriented and multi-camera based intelligent surveillance algorithms and applications.
Main topics are, but are not limited to:
We encourage paper submissions on the followings topics include but are not limited to:
Surveillance benchmark/datasets and performance evaluation
Surveillance object detection, recognition and location;
Surveillance object indexing and retrieval;
Multiple object/subject motion tracking;
Object (face, pedestrian) identification and re-identification;
Multi-modal re-identification
Action/activity/event/behavior recognition;
Large spatio-temporal event recognition;
Object and human re-identification under distributed camera system;
Multi-task recognition;
System and Performance Evaluation;
Other Related Topics